migrate 1.9 openv container

bread-baker

Member
Mar 6, 2010
432
0
16
ti did this and it almost works except the .mount file:

our hostname is fbc158 so,

set a link so .conf and .mount get to correct place:
Code:
ln -s /etc/pve/nodes/fbc158/openvz /etc/vz/conf

restore the backup
Code:
root@fbc158 /var/lib/vz/fbc6bkp # vzrestore vzdump-openvz-101-2011_09_24-23_00_02.tar 3401
...tar: write error
INFO: restore openvz backup 'vzdump-openvz-101-2011_09_24-23_00_02.tar' successful

I'm not sure what the 'tar: write error' was . i assume it is related to not being able to set execute mode on the .mount file, but I'll check into it further tomorrow by setting -v on the tar command ?


try to run it
Code:
vzctl start 3401

Starting container ...
Initializing quota ...
Error exec /etc/pve/openvz/3401.mount: Permission denied
Error executing mount script /etc/pve/openvz/3401.mount

but it will start without the .mount file
Code:
mv 3401.mount 3401.mount-

vzctl start 3401


Starting container ...
Container is mounted
Adding IP address(es): 10.100.100.207
Setting CPU units: 50000
Setting CPUs: 4
Container start in progress..


So next thing to figure out is how to set execute permission on the .mount :
Code:
-rw-r----- 1 root www-data  365 Sep 30 22:51 3401.mount
root@fbc158 /etc/pve/nodes/fbc158/openvz # chmod +x 3401.mount
chmod: changing permissions of `3401.mount': Function not implemented

or how are bind mounts done in 2.0 ?
 
Last edited:
So next thing to figure out is how to set execure permission on the .mount :

It is not possible to set exec permission on the cluster fs - I guess we need to tweak the
vzctl binary/scrips a bit so that it does not require that.
 
I have uploaded a fix which automatically sets the exec permission for openvz action scripts. Please can you test if you can now start the container?
 
I have uploaded a fix which automatically sets the exec permission for openvz action scripts. Please can you test if you can now start the container?

all set now:
Code:
root@homenet-fbc158 /bkup/fbc6bkp # vzrestore vzdump-openvz-169-2011_09_24-23_14_53.tar 3169
INFO: restore openvz backup 'vzdump-openvz-169-2011_09_24-23_14_53.tar' using ID 3169
INFO: extracting archive 'vzdump-openvz-169-2011_09_24-23_14_53.tar'
tar: write error
INFO: Total bytes read: 488222720 (466MiB, 44MiB/s)
INFO: extracting configuration to '/etc/vz/conf/3169.conf'
INFO: restore openvz backup 'vzdump-openvz-169-2011_09_24-23_14_53.tar' successful
root@homenet-fbc158 /bkup/fbc6bkp # ll /etc/vz/conf/3169*
-rw-r----- 1 root www-data 1664 Oct  3 06:24 /etc/vz/conf/3169.conf
-rwxr----- 1 root www-data  310 Oct  3 06:24 /etc/vz/conf/3169.mount
root@homenet-fbc158 /bkup/fbc6bkp # vi /etc/vz/conf/3169.mount
root@homenet-fbc158 /bkup/fbc6bkp # vzctl start 3169
Starting container ...
Initializing quota ...
Container is mounted
Setting CPU units: 1000
Setting CPUs: 1
Configure veth devices: veth169.0 veth169.1 
Adding interface veth169.0 to bridge vmbr0 on CT0 for CT3169
Adding interface veth169.1 to bridge vmbr1 on CT0 for CT3169
Container start in progress...

there is still : tar: write error

if there is something I can do to to help track that down, let me know.
 
So I must use git://git.proxmox.com/git/ for the path to each sub module. Many thanks!
...
Looking in the vzdump repo, I don't see the above mentioned changes, and there are no other remote branches, so I'll assume you have a private area to exchange scripts/diffs with other devs. Until I figure the locations out, I'll just manually rsync things and put them where they need to go.

Again thanks :)
 
No, that assumption is totally wrong ;-) You just need to look at the correct module (pve-cluster):

https://git.proxmox.com/?p=pve-cluster.git;a=commitdiff;h=fd754ca16616f0fadd75899e93192a14ca026027
Awesome, thanks once again. Two notes:
1. I had to force autoconf1.10 because 1.9 (1.96) barfed on @MKDIR_P@
2. I had to manually install lintian for final step of make to succeed

Other than that, it built properly (along with dev prereqs libqb and corosync). Looking forward to the rest of the improvements!
 
what package caused that error?
On a 2.0 Beta install, I install build-essential, which pulled in a bunch of packages. Then I installed all of the suggested packages (which included "autoconf") except for a couple of doc packages that apt-get couldn't find. Once done with that, I installed git and cloned pve-cluster, corosync-pve, and libqb. I attempted to build pve-cluster and libqb, but both required some additional dev packages so I installed them. I then built pve-cluster and libqb by running make and installed the resulting dev packages.

Then I tried to build pve-cluster and it failed at the very end, complaining about not being able to find @MKDIR_P@ in the path. Obviously, something wasn't expanding right from the Makefile.in. Whether it is something I did wrong or out of sequence or if it's a bug, I am not 100% sure. A quick google led me to believe that versions of autoconf 1.9 and below had issues with mkdir detection, so I installed autoconf1.10; make clean && make then ran successfully.

I apologize, but I have already blown away this 2.0 installation and installed 1.9, but that had some issues migrating from our 1.8 machine, so I blew THAT away and installed 1.8, and now live migration works again. I'm willing to try just about anything, so let me know.
 

About

The Proxmox community has been around for many years and offers help and support for Proxmox VE, Proxmox Backup Server, and Proxmox Mail Gateway.
We think our community is one of the best thanks to people like you!

Get your subscription!

The Proxmox team works very hard to make sure you are running the best software and getting stable updates and security enhancements, as well as quick enterprise support. Tens of thousands of happy customers have a Proxmox subscription. Get yours easily in our online shop.

Buy now!